VEHICLE NO X AND PM1 EMISSION FACTORS FROM SYDNEY S M5-EAST TUNNEL Mark F. Hibberd 1 1 CSIRO Atmopheric Reearch, PB 1, Apendale, Vic 3195, Autralia Abtract Reult are preented on the vehicle emiion factor for NO x and PM1 for vehicle uing the M5-Eat tunnel in Sydney. The emiion factor are derived from a tatitical analyi of routine in-tack pollution monitoring data and hourly vehicle count. Vehicle are plit into light petrol vehicle (LPV) and heavy dieel vehicle (HDV) with eparate emiion factor derived for each for both free-flowing and congeted traffic condition. NO x emiion factor agree well with other tudie (.92 g/km for LPV and 15. g/km for HDV) with no ignificant influence from congetion. In contrat, the PM1 emiion factor (.1 g/km for LPV and 2. g/km for HDV) demontrate a trong influence of both congetion and road grade. Firt, the HDV emiion factor i at leat three time larger than reported elewhere (uually.4.6 g/km). Thi larger value i confirmed by a value of 1.6 g/km for Melbourne Burnley tunnel derived from a preliminary analyi of publihed data for congeted condition. Secondly, the LPV emiion factor depend on congetion and i a factor of four higher on weekday and in congeted condition (.22) than on the weekend or in free-flowing traffic (.5 g/km). Keyword: Road tunnel, light petrol vehicle, heavy dieel vehicle, ventilation, fleet average, congetion, CityLink, Melbourne 1. Introduction The amount of pollution emitted from road traffic vehicle i uually decribed uing emiion factor, which pecify the ma of a pollutant emitted per vehicle per unit ditance travelled (unit of g/km). A number of technique are ued to determine emiion factor including dynamometer and on-road teting of individual vehicle a well a fleet-averaged meaurement through ampling near road or in road tunnel. In city road tunnel with ventilation ytem deigned to minimie or eliminate portal emiion, all the pollution i removed via one or more vent tack. Thi enable vent tack pollution monitoring data to be ued together with tunnel vehicle count to compute fleet-averaged vehicle emiion factor. If the vehicle uage data ditinguihe between different vehicle type (commonly length), emiion factor can be determined for each of thee vehicle type. Tran et al (23) adopted thi technique for the 1.6 km long Domain tunnel in Melbourne (part of the City Link network) and extended it uing caniter ampling to VOC (volatile organic compound), which are not regularly monitored in the tack. The reult highlighted ignificant difference from Melbourne emiion inventory. In thi paper we report on emiion factor derived from Sydney M5 tunnel, which i located between Bexley North and Arncliffe. 2. Input Data 2.1. Tunnel ventilation ytem The M5-Eat tunnel conit of two 4-km long twolane tube, one carrying eat-bound traffic, the other wet-bound traffic. The ventilation ytem involve the circulation of air between the tunnel. Freh air i upplied near the centre of each tunnel. Thi i aited by jetfan to move at approximately 3 km/hr in the direction of travel of the vehicle toward the portal. Near the exit portal of each tunnel there i a fan-aited cro-connection to tranfer thi air to the other tunnel. Here it i again aited by jetfan to move with the traffic toward the centre of the tunnel, where it i exhauted through a vent, which i located jut uptream from the freh air upply for that tunnel. Large exhaut fan vent thee gae to the atmophere through a 35-m high vent tack at Turella.
2.2. Sampling period Data were available for five one-week period during September 22 to February 23. Quality control check eliminated about half thee data due to miing data, unuual traffic pattern, error in the tack emiion data, or extremely elevated ambient pollution level due to buhfire. Thi left 18 day of reliable data (8 weekday, 4 Saturday and 4 Sunday/public holiday). The elected day in 22 were: September 5 11; October 7, 12 & 13; November 16 17; December 6, and in 23: February 1, 3 6. 2.3. Vehicle count Vehicle uing the M5 tunnel are counted a hort (< 6 m), medium (6 12 m) or long (> 12 m) vehicle. According to Coffey Geocience (23), 95% of paenger VKT (vehicle kilometre travelled) in Autralia in 2 were in petrol-driven vehicle and 99 1% of long vehicle were dieel-driven. For the purpoe of thi analyi, vehicle in the three length clae are divided into light petrol vehicle (LPV) and heavy dieel vehicle (HDV). All hort vehicle and 75% of medium length vehicle are taken to be LPV with equal pollution emiion factor. All long vehicle and 25% of medium vehicle are taken to be HDV with equal emiion factor. The 75%:25% plit of medium length vehicle into petrol:dieel i baed on an analyi of expected traffic conducted in the planning tage of the M5 (Hyder 2) and i imilar to the VKT plit of 73:23:4 for petrol:dieel:lpg reported by Coffey Geocience (23) for light commercial vehicle. The plit ha alo been confirmed for the current data et by ome analyi carried out a part of thi tudy uing the three length-baed categorie. (However, thee reult are not preented here becaue of the large uncertaintie in the emiion factor.) Figure 1 how the average of the total hourly count on all lane of the tunnel of the two vehicle type for weekday, Saturday and Sunday. The figure how the morning and afternoon peak in the weekday number of LPV a well a the imilar number of LPV around midday on all day of the week. The proportion of HDV i greatet during the middle of weekday (9%) and lowet on the weekend after midday Saturday (about 1%). Average total number of vehicle/hour 6 4 2 Light petrol vehicle Heavy dieel vehicle 6 12 18 6 12 18 6 12 18 24 Weekday Saturday Sunday Figure 1. Average hourly vehicle count of LPV (mainly car) and HDV (mainly truck) uing the M5-Eat tunnel (both direction, all lane). 2.4. Stack emiion The hourly emiion rate of pollutant from the vent tack i calculated from meaured quantitie a: E = c V (1) where the ubcript refer to the tack. E i the emiion rate of a particular pollutant from the tack (unit of g/h), c i the concentration of that pollutant in the tack exhaut (g/m 3 ), and V i the volume flow rate of air out of the tack (m 3 /h). Figure 2 how the average NO x emiion rate from the tack for the three type of day a well a the tandard deviation, which i typically.5 1 g/. The imilar graph for PM1 in Figure 3 how a greater reduction from weekday to weekend emiion, particularly on Sunday, compared to Figure 2, which indicate that the ratio of HDV to LPV emiion rate i greater for PM1 than for NO x (a hown later). A further intereting feature of the figure i the relatively contant weekday tack emiion rate from 6 to 16 hour compared to the trong variation in vehicle number during thi period (Figure 1). Stack NOx emiion rate (g -1 ) 1 5 6 12 18 6 12 18 6 12 18 24 Weekday Saturday Sunday Figure 2. Average and tandard deviation of NO x emiion rate from M5-Eat vent tack.
Stack PM1 emiion rate (g -1 ) 1.5 6 12 18 6 12 18 6 12 18 24 Weekday Saturday Sunday Figure 3. Average and tandard deviation of PM1 emiion rate from M5-Eat vent tack. 3. Analyi procedure Auming that: there are no leakage of emiion from the portal, all emiion are from motor vehicle, and all intrument ued to meaure volume flow rate, pollutant concentration and vehicle number are properly calibrated, then a ma balance of the emiion in the tunnel taking into account the pollution level in the ambient air lead to an equation for the tack emiion rate: E ( nlpv elpv + nhdv ehdv ) L + cb V = (2) where n i the number of vehicle per hour, e i the vehicle emiion factor (g/km), L i the length of the tunnel, and c b i the concentration of the pollutant in the (background) air that i drawn into the tunnel. Hourly background concentration ued in Eq. (2) were obtained from the nearby ambient air quality monitoring tation at Bardwell Valley. The average background concentration of NO x and PM1 were 43 µg/m 3 and 21 µg/m 3, repectively. A linear regreion procedure wa ued to olve for the unknown emiion rate e LPV and e HDV. 4. Reult The data were firt analyed with all day of the week included. The calculated emiion factor are hown in the firt row of Table 1. Separate analye were then carried out for weekday, Saturday, and Sunday data. The only ignificant difference found were in the PM1 emiion factor for LPV, which were a factor of four lower on the weekend than on weekday. Further invetigation revealed that thi difference wa related to the difference in average peed and hence to congetion level. Average vehicle peed (km/hr) 8 7 6 5 4 2 4 6 Total vehicle/hour Weekday Weekend Figure 4. Speed v. vehicle count howing difference between weekday and weekend traffic. Figure 4 how the average peed veru the hourly vehicle count for all hour ued in thi analyi. Weekday data are ditinguihed from weekend data uing different ymbol. The weekend data overlap much of the weekday data and it i ignificant that all weekend data correpond to average vehicle peed greater than 65 km/hr and hourly vehicle count le than 5. The circled data how the weekday data that are clearly different from the weekend data. Thee correpond to the highet vehicle count with the lowet vehicle peed, indicating congeted condition. They alo correpond to the highet tack emiion rate. Some caution need to be exercied in interpreting the average peed becaue they are a weighted average of all vehicle in both tube of the tunnel, one of which may be congeted while the other i free-flowing. The emiion factor for average vehicle peed above and below 65 km/hr are lited in Table 1. Except for the PM1 emiion factor for LPV, there are no tatitically ignificant difference Table 1. Emiion factor computed from the M5-Eat data for light petrol vehicle (LPV) and heavy dieel vehicle (HDV) and comparion with ome other data. Condition NO x emiion factor (g/km) PM1 emiion factor (g/km) LPV HDV LPV HDV
M5 tunnel, All peed.92 ±.2 15. ±.4.1 ±.2 2. ±.3 M5, < 65 km/hr (& weekday).89 ±.6 14.4 ± 1.2.22 ±.5 1.98 ±.1 M5, > 65 km/hr (& weekend).96 ±.2 15.5 ±.3.5 ±.1 1.93 ±.3 CityLink, 2 (weekday).82 ±.4 16.7 ±.3.9 ±.1.615 ±.11 NSW EPA, 23; arterial road 1.1 (congeted +1%) 12 (congeted +25%).28 (congeted +7%).3.5 (congeted +7%) between the all peed reult and thoe eparated according to peed. For LPV, the PM1 emiion factor for peed le than 65 km/hr i.22 g/km (the ame a that obtained from analyi of all weekday data), wherea it i.5 g/km for peed greater than 65 km/hr (the ame a obtained from an analyi of all weekend data). Thi difference in LPV PM1 emiion factor appear urpriingly large, but it hould be noted that the PM1 meaured in the vent tack include not only tailpipe emiion, but alo contribution from tyre and brake wear and re-upended road dut. The NPI (2) manual for etimating motor vehicle emiion lit the PM1 contribution from brake wear a.8 g/km and from tyre wear a.5 g/km for paenger vehicle. Thee are derived from brake and tyre lifetime average, and o depend on traffic and road condition. Thee ource provide a poible explanation for the difference of.17 g/km between the free-flowing and congeted PM1 emiion factor; it i potulated that congeted condition have a much greater contribution from brake and tyre wear. A more detailed chemical analyi of the particle emiion from the vent tack would be needed to confirm thi hypothei. Thi effect i unlikely to be detectable in HDV emiion factor becaue the tailpipe emiion are approximately 1 time larger and the uncertaintie are greater than the above-mentioned brake and tyre wear contribution. 4.1. Verification of computed factor In order to check the reult of the tatitical analyi, the meaured tack emiion rate are compared with thoe derived uing the computed emiion factor. Figure 5 how very good agreement between the meaured and computed NO x emiion rate. The abolute difference are le than 1 g/ and the fractional difference are generally maller than 5%, except between 22 and 2 hour, when there are difference of up to 2%. The figure alo how that petrol and dieel Stack NO x emiion rate (g/) 15 1 5 Meaurement Modelled total LPV (.92 g/km) HDV (15 g/km) bkgnd NOx 6 12 18 24 Time of day (weekday) Figure 5. Comparion between modelled and meaured NOx tack emiion rate uing emiion rate derived in thi paper. vehicle each contribute about half the total emiion for mot of the day. The background contribution, repreenting the lat term in Equation (2), i generally le than 1%. The trough in LPV contribution during the middle of the day i filled in by the increaed HDV emiion to produce approximately contant NO x tack emiion from 6 to 16 hour. Stack PM1 emiion rate (g -1 ) 1.4 1.2 1.8.6.4.2 Meaurement Modelled total LPV (.1 g/km) HDV (2. g/km) bkgnd PM1 6 12 18 24 Time of day (weekday) Figure 6. Comparion between modelled and meaured PM1 tack emiion rate uing emiion rate derived in thi paper.
Stack PM1 emiion rate (g -1 ) 1.4 1.2 1.8.6.4.2 Meaurement Modelled total LPV (.5 &.22 g/km) HDV (2. g/km) bkgnd PM1 6 12 18 24 Time of day (weekday) Figure 7. A for Figure 6 but with LPV rate elected according to average vehicle peed, a decribed in the text. For PM1, two comparion are hown, one (Figure 6) uing the all peed emiion factor for LPV of.1 g/km, the other (Figure 7) uing the two emiion factor of.5 g/km (for vehicle peed greater than 65 km/hr) and.22 g/km (le than 65 km/hr). Figure 7 how a lightly better agreement with the meaurement, mainly during daylight hour, but becaue of the dominance of the HDV contribution, the light difference in the LPV emiion factor have almot no effect on the maximum daytime tack emiion rate. The meaured and computed total daily PM1 tack emiion ( 4 kg/day) agree to within.4% (Figure 6) and 1.9% (Figure 7) but the figure how that between 22 and 4 hour the computed emiion rate conitently over-etimate the tack data by about.5 g/. Thee time correpond to low traffic volume and period when the exhaut fan feeding the tack are operating at about half the daytime rate of 93 m 3 /. In addition to thee night-time over-etimate, there are alo under-etimate by about.1 g/ at 6, 15 and 16 hour. Although thee dicrepancie do not affect the maximum daytime emiion rate of about.9 g/, they point to more complexitie in the PM1 tory. 4.2. Comparion with other reult Table 1 alo compare thee reult with emiion factor reported elewhere for the Autralian vehicle fleet. Tran et al (23) reported data from the CityLink Domain tunnel in Melbourne. The NSW EPA ha an ongoing program for etimating emiion factor; the lited value are for arterial road and are recommended for regional modelling. The NO x emiion factor from thi tudy agree within 1 15% with the CityLink reult and within 25% with the NSW EPA figure. The cloe agreement between the CityLink and M5 tunnel reult ugget that the current NSW EPA emiion factor may be 25% low for HDV and 2% high for LPV. The aymmetry of the dicrepancie mean that it i unlikely that the difference of the NSW EPA factor are due to difference in vehicle operating condition in the two tunnel. It i alo unlikely that there are ignificant diparitie between the vehicle fleet uing the two tunnel. In contrat to NO x, there are large difference between the PM1 emiion factor. The weekday reult from thi tudy are 2.5 3 time larger than thoe reported for the CityLink tunnel. A potentially ignificant difference between the tunnel i that the Domain tunnel ha a relatively flat grade with jut a hort climb near the exit, wherea the M5 tunnel ha 4 m of climb at 5 8% grade out of the tunnel at each end (Figure 8) and at the exit ramp at Marh St and the Prince Highway. 5% 4 m.5% 13 m.5% 17 m Figure 8. Grade along the M5 tunnel. 8.3% 25 m 2.5% 15 m PIARC (1995) report an increae of approximately 2% in emiion factor at 7 km/hr for each 1% increae in road grade for NO x and PM1 (the latter meaured a turbidity). While thee increae are conitent with the reult from thi tudy for NO x, the PM1 reult indicate a much tronger, and poibly non-linear, dependence on road grade. Similarly, there i a tronger effect of congetion on PM1 emiion factor ued by the NSW EPA, namely 7% for PM1 compared to 1 25% for NO x. An etimate can be made of the enhancement due to road grade required to achieve the three-fold increae in PM1 emiion factor in the M5 tunnel. If the influence of road grade on emiion factor i linear and downhill grade are aumed not to reduce emiion, then the equivalent grade through the whole M5 tunnel i.7.8%. If the three-fold increae in the PM1 emiion factor were all due to road grade, then thi would be equivalent to a 3 4% increae for a 1% grade. Thi i a very large enhancement and need verification before being ued for etimating emiion elewhere. 4.3. Peak emiion from CityLink tack A review of data from the Melbourne CityLink tunnel (EPAV 24) provide ome peak tack emiion data for the Domain and Burnley tunnel, each of which carrie up to three lane of traffic in one direction. (The Domain tunnel i decribed above; the 3.4 km long Burnley tunnel include a 1 km climb at 5% grade to the exit portal.) The
monthly maximum hourly-average tack emiion from March 22 to February 24 are typically.21 g/ from the Domain tunnel and.66 g/ from the Burnley tunnel (EPAV, 24). A quick etimate of the expected tack emiion uing the emiion factor lited in Table 1 can be made if maximum hourly vehicle number are known, but thee are not readily available. For thi analyi we aume that peak traffic flow are the ame in both the Burnley and Domain tunnel. We cale up the peak Tran et al (23) number for the year 2 by 2% to account for increaing traffic volume. Thi give 4 LPV/hour and 4 HDV/hour with an etimated uncertainty of about ±2%. Thee flow are not inconitent with peak traffic flow per lane reported by PIARC (1995) for uni-directional urban tunnel of 2 car/hour in free-flowing and 1 car/hour in congeted traffic. Table 2 lit the meaured and computed peak tack emiion uing emiion factor for the M5 tunnel a well a thoe from Tran et al. The apparent inconitency between the Tran et al reult for the Domain tunnel and thoe lited in Table 2 i due to thee tack emiion rate being the highet each month, i.e. correponding to more congeted condition than the average conidered by Tran et al. (ee below). Available data (e.g Figure 7) indicate that about 9% of peak weekday PM1 tack emiion are due to HDV. Uing thi fact, we can cale the reult in Table 2 to derive the PM1 HDV emiion factor for the Domain tunnel to be approximately 1. g/km and for the Burnley tunnel approximately 1.6 g/km. Table 2. Comparion of meaured and modelled peak tack emiion rate for Melbourne City Link tunnel. Tunnel Domain Burnley Length 1.62 km 3.4 km Meaured maximum tack emiion rate (EPAV, 24) Thi paper emiion factor LPV.1 g/km HDV 2. g/km Tran et al (23) emiion factor LPV.9 g/km HDV.615 g/km Etimated HDV PM1 emiion factor.21 g/.66 g/.38 g/.79 g/.13 g/.27 g/ 1. g/km 1.6 g/km Thi HDV emiion factor of 1. g/km for the Domain tunnel i 6% larger than the Tran et al value of.615 g/km, but thi i conitent with the 7% increae in PM1 emiion factor uggeted by NSW EPA for congeted condition. The etimated emiion factor for the Burnley tunnel in congeted condition of 1.6 g/km i halfway between that for the Domain tunnel and the M5 tunnel. Thi reult upport the emiion factor of 2. g/km derived above for the M5 tunnel and ugget that a combination of congetion and uphill road grade doe in fact produce large increae in the PM1 emiion factor. 5. Concluion Thi paper preent vehicle emiion factor for PM1 and NO x derived from a tatitical analyi of routine in-tack pollution monitoring data and hourly vehicle count from Sydney M5-Eat tunnel. The derived NO x emiion factor are.92 g/km for light petrol vehicle (LPV) and 15. g/km for heavy dieel vehicle (HDV). Thee number are not affected by congetion level and agree within 25% with reult from other Autralian tudie. The derived average PM1 emiion factor are.1 g/km for LPV and 2. g/km for HDV. In contrat to the NO x figure, thee reult may be controverial. Firt, the PM1 HDV emiion factor i three time larger than reported elewhere, including by Tran et al (23) for Melbourne Domain tunnel. However, a quick analyi uing peak tack emiion data correponding to congeted condition in Melbourne tunnel (EPAV 24) and reaonable aumption about the traffic volume produce an HDV emiion factor for the Domain tunnel of 1. g/km, which i conitent with a 7% enhancement due to congetion (a ued by NSW EAP). Thi analyi alo give an HDV emiion factor for Melbourne Burnley tunnel of 1.6 g/km, which i halfway between that for the Domain and M5 tunnel. Secondly, the PM1 LPV emiion factor depend on the level of traffic congetion. It i found to be a factor of four higher in congeted traffic (.22 g/km) than in free-flowing condition (.5 g/km). Thirdly, although the meaured total daily PM1 tack emiion rate ( 4 kg/day) agree with that computed uing the derived emiion rate to within about 1%, the computed hourly emiion rate conitently over-etimate the tack data by.5 g/ at night (22 4 hour) and underetimate it by about.1 g/ at 6, 15 and 16 hour. Although thee dicrepancie do not affect the maximum tack emiion rate of about.9 g/, they point to more complexitie in the PM1 tory.
Obviouly more work i needed to eparate out the influence of congetion and road grade on PM1 emiion factor. Acknowledgment The data ued in thi tudy have been kindly upplied by NSW Health, the community group RAPS (Reident Againt Polluting Stack), and Charle Xu (NSW EPA). Reference Coffey Geocience, 23. Fuel Quality and Vehicle Emiion Standard Cot Benefit Analyi. Available at: www.ephc.gov.au/ltec/pdf/ CBAFinalReport23October.pdf EPAV, 24. CityLink Review of Air Quality Modelling, EPA Victoria Publication 958, available at: http://epanote2.epa.vic.gov.au/epa/publication. nf Hyder, 2. M5 Eat Motorway driven tunnel ervice mainline tunnel air quality (35 m tack). Document #WCR2, reviion B. Report by Hyder Conulting. NPI, 2. Emiion Etimation Technique Manual for Aggregated Emiion from Motor Vehicle, National Pollution Inventory, 22 November 2 Verion 1., http://www.npi. gov.au/handbook/aedmanual/pub/motorvehicl e.pdf PIARC, 1995. Vehicle Emiion, Air Demand, Environment, Longitudinal Ventilation, Committee on Road Tunnel, XXth World Road Congre, Montreal, ISBN 2-846-34-X Tran, T.V., Ng, Y.L. and Denion, L., 23, Emiion factor for in-ervice vehicle uing Citylink tunnel, in: Proceeding of the National Clean Air Conference (on CD), Newcatle, Clean Air Soc Aut & NZ.